Request a briefing

Markings, computed.

Marque checks classification markings during live work or automated batch processing. Its patent-pending rule resolution system computes a correct result for each marking and all of them together.

You'll never have to write a banner marking again, or spend minutes figuring out what it should include.

DETERMINISTICRULE-DRIVENLIVE OR BATCHFREE TEXTAUTO-FIXESOPTIONAL

How it resolves

What the rules determine.

Marque identifies markings in raw text or structured data and evaluates them together against their governing rules. When the rules determine one result, Marque returns it. When the supplied facts allow several compliant results, Marque returns the valid options and explains what distinguishes them.

Each result comes with a plain-language explanation and its supporting rules. When the available facts do not determine one correction, Marque identifies the remaining question and gives the classifier the information needed to answer it.

Banner marking1 rule applied
SECRET//REL TO GBR
SECRET//REL TO USA, GBRDERIVED
CAPCO-2016 §H.8 p151REL TO requires USA in the country list; Marque adds it as a rule-derived correction.
CONSTRUCTED EXAMPLEIt's unclassified. About this example.

Where Marque fits

Where intent becomes a marking.

Marque fits wherever classification intent becomes a marking: during live work, across a collection, or downstream of another system. Its inputs come from a classifier, existing documents, or AI and natural-language-processing tools.

The responsible authority or accredited upstream tool owns the semantic classification determination. Marque checks and corrects the portion markings, banners, labels, and metadata that express that determination, bringing them into compliance with the governing marking rules.

The division of work

What Marque does—and what stays upstream.

Marque decides

Rule-bound marking outcomes

Marque determines whether a marking is valid, how banner and page roll-ups resolve, which classification floors and dissemination constraints apply, and whether a declassification form is permitted.

Eligible corrections

When the rules yield an eligible determined or ranked correction, Marque can apply it under the configured promotion controls. Unranked alternatives never auto-apply.

Upstream authority

The semantic classification determination

Whether source material warrants classification comes from the original classification authority or another authorized upstream process. Marque does not infer that determination from the subject matter.

The controlling facts supplied upstream

A person or upstream system supplies the applicable classification and SCG facts. Marque checks those inputs and automates the marking consequences the governing grammar determines.

Prevent damaging spills
Prevent, detect, and correct common human errors like an overlooked TOP SECRET portion on page 32 in a document marked SECRET. Or more nuanced situations, like when a portion is marked (C//TK) in a document labeled CONFIDENTIALbut TK must be at least SECRET

How it deploys

Local integration

Marque runs inside an application, browser, or web app, or directly on the command line.

REST API and batch processing

Marque integrates through a REST API or runs as a deterministic stage in automated batch and extract-transform-load (ETL) pipelines.

Deployment stays under customer control

Marque does not phone home. The customer controls network isolation, audit routing, persistence, retention, and access.

Marque keeps checking, correction, and write-back separate. The caller chooses which step to request and whether returned text replaces the source. Marque returns machine-readable records when the caller reports a fix as applied and when the available inputs do not determine one correction.

Finding errors adds work. Correcting them removes it.

Marque is sold through resellers, distributors, and integrators.

Briefings demonstrate Marque against agreed test inputs.